stefan
Anyone trying to follow this will need to reference the correctly formatted question:
https://math.stackexchange.com…uniform-geodesics#2316201
I have a problem with your question, which is that I don't understand how you define the mapping F:
Furthermore if we take F:S2←P(I×S1)F:S2←P(I×S1), P(⋅)P(⋅) the power set, as the mapping of a point of the sphere towards a discrete point of the geodesics covering that point, a function that we constrain G so that it exists.
F would appear to be a function domain I X S1 (the set of all points on all great circles) and range S2 (the surface of a sphere embedded in R3). I think you need the arrow in the opposite direction so that with F you are mapping a point on S2 to the set of all great circles that go through that point? I don't understand how this constrains G because it will exist (though may be trivial) for all G. I also expect that you actually require the existence of the Euclidean metric on R3 which induces a manifold on the embedded S2 - because you assume differential structure below.
The constraint on Mu is central to this definition:
Then we also constrain the measure μμ to satisfy ∀p∈S2 :(∑β∈F(p)dμ(β)=1/(4π)dS2∀p∈S2 :(∑β∈F(p)dμ(β)=1/(4π)dS2
Mu is defined to be a positive measure on I X S1. S1 is embedded in R2 and inherits a manifold structure from that embedding. I however is a set with no metric structure. That is OK, the fact that Mu is a measure will impose a measure (but not metric nor even topology) on I X S1. I think however you want this measure to be compatible with the implicit embedded local metric on S1? And maybe (see below) you want some extra structure on I.
Beta is the set of all great circles through p and is a subset of P(I X S1). Mu(beta) is therefore the measure on this subset. I think you are actually using the induced manifold structure of S2 here (p in S2) and saying in this constraint that the measure must be locally symmetric wrt any rotation of S2 (basically, small patches on S2 with the same area will have the same measure).
For this to be a proper question we need it to be expressed much more tightly. Also, I suspect that we don't need S1. The whole problem becomes simpler to think about if you just ask questions about the nalpha. For example the set of nalpha corresponding to great circles going through a point p is makes the great circle whose normal is itself p. A pleasing and well-known symmetry.
Now I'm still unclear about the symmetry here induced on G. The obvious symmetry satisfying this condition leads to Total = 0. It would help elucidate this to show constructively (without introducing concepts from physics like angular momentum) the existence of any non-trivial solution with Total not equal to 0. You just need to give the solution as math and show it complies.
When you have tied this up I still doubt your result is true, unless you have the full euclidean induced manifold structure on S2 and something similar on I. After all the hand-waving physical model you wish to hook to this question certainly has I isomorphic to S2 and a natural euclidean metric (that is - the set of great circles on S2 is isomorphic to their normals, which are isomorphic to S2. In fact this isomorphism makes a duality). I also expect (but don't know) that it can be shown true much more trivially with a much simpler isomorphic structure than you have set up here. I'm uneasy about mixing a measure with a manifold which seems plain weird.
You can't ask people to think about this without some more work on what this constraints really means. This, together with the above loose points in the definition, make for the bad rating I believe.
Display More
Thanks, this is good feedback.
So what I'm struggling to say rigorously is:
We consider a subset of all geodesics and specifically a subset where only a discrete number of selected geodesics goes through a point p on the sphere, so for a point, there is a selection of geodesics, {S_a_1,S_a_2,...} and for each of the geodesics you have a point which cover p, e.g. you have pairs {(a_1,p_1),(a_2,p_2),...} eg a subset of (I x S^1) e.g. en element of the powerset now each of these points has a measures or if you like infinetismal weights mu(a_1,p_1), mu(a_2,p_2),... and they should sum up to 1/4pi dS. and then you can integrate the surface and recover total mass of 1. So this is a fancy way of saying that the sum of the coverings is uniform. In order for there to be a covering of type G I assume that this mapping F should exists and it does in Mills example (but it is hidden). Your good point is that perhaps I must better motivate the existance of F.
> Mu is defined to be a positive measure on I X S1. S1 is embedded in R2 and inherits a manifold structure from that embedding. I however is a set with no metric structure. That is OK, the fact that Mu is a measure will impose a measure (but not
> metric nor even topology) on I X S1. I think however you want this measure to be compatible with the implicit embedded local metric on S1? And maybe (see below) you want some extra structure on I.
I am sloppy or rusty whatever you like to call it and do find it tricky to get the formulation right, I understand your points here.
> Beta is the set of all great circles through p and is a subset of P(I X S1). Mu(beta) is therefore the measure on this subset. I think you are actually using the induced manifold structure of S2 here (p in S2) and saying in this constraint that the measure > must be locally symmetric wrt any rotation of S2 (basically, small patches on S2 with the same area will have the same measure).
Yes
> For this to be a proper question we need it to be expressed much more tightly. Also, I suspect that we don't need S1. The whole problem becomes simpler to think about if you just ask questions about the nalpha. For example the set of
> nalpha corresponding to great circles going through a point p is makes the great circle whose normal is itself p. A pleasing and well-known symmetry.
> Now I'm still unclear about the symmetry here induced on G. The obvious symmetry satisfying this condition leads to Total = 0. It would help elucidate this to show constructively (without introducing concepts from physics like angular momentum)
> the existence of any non-trivial solution with Total not equal to 0. You just need to give the solution as math and show it complies.
I tried to indicate an example which the Total is not equal to 0, I indicated the construction and it is included in Mills text. Shall I add page after page with a proper deduction? Can't I refer to his book?
> When you have tied this up I still doubt your result is true, unless you have the full euclidean induced manifold structure on S2 and something similar on I. After all the hand-waving physical model you wish to hook to this question certainly has I
> isomorphic to S2 and a natural euclidean metric (that is - the set of great circles on S2 is isomorphic to their normals, which are isomorphic to S2. In fact this isomorphism makes a duality). I also expect (but don't know) that it can be shown true
> much more trivially with a much simpler isomorphic structure than you have set up here. I'm uneasy about mixing a measure with a manifold which seems plain weird.
The generality of measures is perhaps too much, perhaps I should be working with manifolds as you say.
What about this. We want that for all measurable set A on S^2, we have a set F(A) \in P(I x S^1) so that F(A) is measurable with mu and mu(F(A)) = n(A), n being the
uniform measure on S^2?